[freehaven-cvs] cleanup throughout

[arma@seul.org (Roger Dingledine)]

Update of /home/freehaven/cvsroot/doc/sync-batching
In directory moria.mit.edu:/home2/arma/work/freehaven/doc/sync-batching

Modified Files:
	sync-batching.tex 
Log Message:
cleanup throughout


Index: sync-batching.tex
===================================================================
RCS file: /home/freehaven/cvsroot/doc/sync-batching/sync-batching.tex,v
retrieving revision 1.13
retrieving revision 1.14
diff -u -d -r1.13 -r1.14
--- sync-batching.tex	21 Jan 2004 20:40:27 -0000	1.13
+++ sync-batching.tex	21 Jan 2004 21:32:31 -0000	1.14
@@ -39,8 +39,8 @@
 the network. We show that a synchronous batching strategy can be used
 in various topologies including a free-route network (a topology where
 senders choose paths freely), and that such topologies can provide better
-anonymity and better robustness to node failure than a cascade network
-(where users choose from a set of fixed paths).
+anonymity and better message reliability in the face of node failure
+than a cascade network (where users choose from a set of fixed paths).
 
 \end{abstract}
 %======================================================================
@@ -72,13 +72,13 @@
 routes are used.
 \item It allows senders to verify message processing, enabling reputation
 systems such as those described in \cite{mix-acc,casc-rep}.
-\item It helps block blending attacks because mixes cannot delay messages
-without permanently dropping them.
+\item It helps block blending attacks compared to asynchronous batching,
+because mixes cannot delay messages without permanently dropping them.
 % though rgb, but there the adversary can still drop sender messages
 
-* exit policies impose differentiation anyway. might as well make it
-  formally part of the topology.
-* the one-to-a-sender ticket schemes can work
+% * exit policies impose differentiation anyway. might as well make it
+%   formally part of the topology.
+% * the one-to-a-sender ticket schemes can work
 
 \end{tightlist}
 
@@ -86,11 +86,8 @@
 stratified (a restricted route), and free-route. We find that the foo
 topology provides the highest expected anonymity of these three.
 
-%Specifically, whereas
-%with \emph{asynchronous batching} strategies each message is forwarded
-%through the network based on local decisions at each mix, in a synchronous
-%batching design, each batch of messages enters the mix network together,
-%and the messages proceed in lock step 
+
+In Section~\ref{sec:foo} we describe foo, in bar bar, etc.
 
 
 \section{Synchronous batching}
@@ -196,11 +193,12 @@
 synchronous batching. We refer to that paper for a detailed discussion
 of the timing model, how to handle loosely synchronized clocks, and the
 step-by-step instructions for senders and mixes to use the network and
-judge whether messages have arrived in time. The paper also describes
-a receipt and witness system by which senders and mixes can prove that
-a given mix failed to pass on a given message. These receipts allow
-a reputation system where nodes that drop messages lose reputation,
-discouraging misbehavior.
+judge whether messages have arrived in time.
+
+The paper also describes a receipt and witness system by which senders and
+mixes can prove that a given mix failed to pass on a given message. These
+receipts allow a reputation system where nodes that drop messages lose
+reputation, discouraging misbehavior.
 
 \subsection{The Disadvantages of Free Mix Routes}
 \label{subsec:disad}
@@ -222,8 +220,9 @@
 \emph{Eventually, a message is only unobservable in that group of messages
 which have this mix on the same routing position.} But in the synchronous
 design, that's not a problem because this group is large (if only one
-mix is trustworthy, $1/w$ of all messages in the batch, which is what
-it would be anyway). They conclude: \emph{If only one miX of a route is
+mix is trustworthy, $1/w$ of all messages in the batch---which is what
+it would be in a cascade network too). They conclude: \emph{If only one
+mix of a route is
 trustworthy, then the achievable anonymity is distinctly lower in a mix
 network compared to a synchronously working mix cascade.} The actual
 conclusion should be: If only one mix of a route is trustworthy, then
@@ -403,16 +402,23 @@
 
 \subsection{Choose path hops with or without replacement?}
 
-Intuitively, when the adversary density is low, entropy is doing
-fine. When it's high, the difference between picking between $B$ and $B-1$
-bad nodes is slight.
+Conventional wisdom (see e.g.~\cite{disad-free-routes}) suggests that
+in a free-route network, Alice should never choose the same node twice
+in a row when picking her path. The goal is to reduce the chance that
+the path contains only bad nodes. But we find that as the size of the
+network grows, this increased complexity in path selection has little
+impact on Alice's entropy.
+
+Intuitively, when the adversary density is low, entropy will be high
+in either case; whereas when most nodes are owned by the adversary,
+the difference between picking between $B$ and $B-1$ bad nodes is slight.
 
 More formally, if there are $G$ good nodes and $B$ bad nodes, then the
 chance of selecting a bad node as your next link is $\frac{B-1}{G+B}$ if
 the current node is bad and $\frac{B}{G+B}$ if your current node is good.
 The difference between the two is only $\frac{1}{G+B}$. In particular,
-the difference does not depend on what fraction of the nodes are bad. And
-for reasonably large values of $G+B$ (total number of nodes), the shift
+the difference does not depend on what fraction of the nodes are bad. For
+reasonably large values of $G+B$ (total number of nodes), the shift
 in probability distribution is negligible.
 
 \subsection{Robustness}

***********************************************************************
To unsubscribe, send an e-mail to majordomo@seul.org with
unsubscribe freehaven-cvs       in the body. http://freehaven.net/